The single highest-leverage workflow on a service manager's desk in 2026 is the AI-built technician ride-along scorecard. Pre-Rilla, the service manager rode along with a tech for two days, watched eight closes, and gave feedback. Throughput: 2-3 in-person ride-alongs per manager per day. Across a 12-tech team, each tech got coached every 4-6 weeks — annual review cadence dressed up as ride-along discipline. The numbers stayed flat because the coaching cadence could not catch the behavior before it embedded. Rilla broke the throughput ceiling. 30-40 virtual ride-alongs per manager per day on lapel-mic transcripts, scored by AI against the five highest-impact coaching moments, surfaced as a one-page card per tech per day. The service manager reviews 30-40 cards in the time it took to do two in-person ride-alongs before — 10-20x coverage density at $200-$400 per seat per month. This lesson is the architecture of that workflow: the five coaching moments the AI flags, the RGA scoring rubric the manager runs, the daily 15-minute huddle that produces the 18% close-rate lift, the comp redesign that ships at rollout, and the kill rules that surface a Rilla pilot in trouble inside 60 days. The service manager builds this workflow once. The shop runs it for the next decade.

Why the Ride-Along Scorecard Is the Service Manager's Leverage Point

A service manager owns three numbers that move the P&L: technician close rate on service-to-replace pivots, MPR (membership penetration rate), and recall percentage. All three move through the same lever — what happens in the homeowner's kitchen, basement, or attic during the 45-to-90 minute service call. Pre-Rilla, the manager had no observable record of that conversation. The tech recounted it; the homeowner remembered it differently; the close didn't happen; nobody could reconstruct why. Coaching ran on stories, not data.

Rilla converts every service call into a coachable data point. The tech wears a lapel mic on every appointment (not just replacement closes — every call, including diagnostic, repair, maintenance, and warranty). Rilla transcribes the conversation, scores against the five highest-impact moments (intro, system condition explanation, repair-vs-replace pivot, options presentation, financing pivot), aggregates into the RGA (ride-along grade) for that call, and surfaces a coaching card to the service manager by the end of the day. The manager opens 30-40 cards in the morning, picks the bottom-quartile moments to coach, and runs a 15-minute morning huddle with affected techs.

The math: at $620 baseline average ticket × 6 calls per tech per day × 12 techs × 250 days = $5.6M annual service revenue per shop on the service-call side. An 18% close-rate lift on repair-vs-replace pivots — Rilla's documented median across home-services deployments — drives $400K-$800K of incremental replacement revenue annually. MPR lift from a 22% baseline to 38-45% on the same call base adds another $180K-$320K in annual recurring membership revenue. Recall percentage drops as coaching surfaces the diagnostic shortcuts that produced the recall in the first place. Rilla's seat cost across 12 techs at $300/seat/month is $43K annually. Payback measured in weeks.

The Five Coaching Moments the AI Flags Per Ride

Rilla's scoring rubric flags five specific moments on every service call. The five are not arbitrary — they are the five moments where the data shows the close-rate, MPR, and recall numbers actually move. The manager coaches these five; the rest of the conversation is context.

Moment One — Intro and Discovery in the First 90 Seconds

The tech's first 90 seconds at the door determines whether the homeowner is open or guarded for the next 60 minutes. Rilla scores: did the tech use the homeowner's name within the first 30 seconds; did they reference the specific issue from the dispatch notes (not a generic "what brings me out today" opener that primes the homeowner to talk price); did they ask permission to enter the equipment area; did they acknowledge any pet, child, or schedule constraint the homeowner mentioned; did they explain what the diagnostic process will look like and how long it will take. The intro moment scores 0-20 on the RGA composite, weighted 15%. Techs who score 15+ on intro consistently close repair-vs-replace at 8-12 points higher than techs who score 8-10.

Moment Two — System Condition Explanation

The most consequential moment of the diagnostic call. The tech finds what's wrong; how they explain it to the homeowner determines whether the conversation becomes a repair-only transaction or a repair-vs-replace pivot. Rilla scores: did the tech use the homeowner's language (not "the secondary heat exchanger has hairline fractures," but "the part of the furnace that exchanges heat from the burner to your air has cracks you can see"); did they show photos from the tablet (AI-tagged where applicable); did they tie the condition to a homeowner outcome (safety, comfort, monthly utility bill, recurring repair cost); did they avoid the dead-end "you need a new system" close-too-early move that triggers homeowner defensiveness. System condition scores 0-25, weighted 25%. The single highest-impact moment on the RGA.

Moment Three — Repair-vs-Replace Pivot

The pivot is where the close lives on aged equipment. Rilla scores: did the tech present the equipment age + repair cost + future-failure-probability math (the AI surfaces this on the tablet from the equipment record); did they present the partial-replace mid-tier option (often a heat exchanger swap, a coil replacement, a panel upgrade vs. a full system); did they avoid pushing the close too early before the homeowner has internalized the diagnosis; did they invite the homeowner's questions before presenting options. Pivot scores 0-20, weighted 20%. Techs who consistently score 15+ on the pivot lift replace-attach rate from a 22% baseline to 38-45% on systems older than 12 years.

Moment Four — Options Presentation

The good/better/best three-option presentation logic the Comfort Advisor uses on the kitchen-table close lives in compressed form on the service call. Rilla scores: did the tech present three options (repair-only / partial-replace / full-replace); did they distinguish the value proposition of each (efficiency delta, comfort delta, warranty delta, future-failure delta); did they avoid drowning the homeowner in technical specs; did they let the homeowner self-select toward the mid-tier rather than push them. Options scores 0-15, weighted 15%. Mid-tier dominance on the service-call presentation runs 48-55% — slightly lower than the Comfort Advisor's kitchen-table close, but the same cognitive anchoring mechanism.

Moment Five — Financing Pivot

The financing conversation re-anchors the homeowner from the sticker price to the monthly payment. Rilla scores: did the tech mention financing before presenting the sticker (anchoring to payment, not price); did they reference the soft-pull tier (Wisetack for fast service-trade and entry-level replacement, GreenSky for high-ticket replacement, Synchrony for branded revolving lines on multi-system customers); did they handle the spouse-decision-rights conversation if it surfaced; did they hand off to a Comfort Advisor cleanly if the deal sized up beyond the tech's seat. Financing scores 0-20, weighted 25%. The single most under-coached moment in shops without Rilla — techs default to silence on financing because it feels like overstepping; the data shows financing-mentioned calls close at 22-30 points higher.

The RGA Scoring Rubric and How Managers Read It

The RGA (ride-along grade) aggregates the five moments into a single 0-100 score per call. The weighting reflects close-rate impact: intro 15%, system condition 25%, pivot 20%, options 15%, financing 25%. The score is not a grade for grading's sake; it is the leading indicator. Techs at sustained RGA 75+ close repair-vs-replace at 38-45% on aged equipment (systems over 12 years). Techs at sustained RGA 55-65 close at 18-25% — half the rate, on the same call base.

Reading the rubric in practice is a 90-second skim per card. The manager opens the card; sees the RGA number; sees the per-moment scores; sees the AI's two-line callout on the lowest-scoring moment ("system condition explanation: tech used 'secondary heat exchanger' three times without translation; homeowner asked twice what that meant"); and decides whether to coach today or queue for the weekly cohort review. The bottom-quartile moments per tech per week become the coaching agenda. The discipline is moments-coached, not cards-reviewed — 30-40 cards reviewed but 3-5 coaching conversations per day is the cadence that produces the 18% lift.

Cohort tracking is the second-order discipline. Plot every tech's weekly RGA against close rate; the slope tells the manager which techs are coaching-responsive (RGA moves and close rate follows) and which are not (RGA moves but close rate doesn't move, suggesting the tech is performing the coached behaviors mechanically without internalizing them, or the coaching is targeting the wrong moments for that tech's blind spots). For new hires, RGA is the 30/60/90 onboarding ramp: target RGA 50+ at day 30, 65+ at day 60, 75+ at day 90. Techs who don't hit day-90 target are not destined for the close-heavy service seat; the data is honest about that within 3 months instead of 12.

The Daily 15-Minute Morning Huddle That Produces the Lift

Rilla without the daily huddle is a recording. The huddle is the mechanism that converts AI scoring into behavior change. The shops that get the 18% lift run it five mornings a week at 6:45-7:00 a.m. before techs roll. The shops that batch coaching to Friday afternoons get 8-10% lift — about half — because the feedback decay between observation and coaching damages the rep's ability to internalize the moment.

The huddle structure is tight. Two minutes per tech maximum; rotate which tech leads the recap. The manager opens with the previous day's bottom-quartile moment for one tech: "Marco, your 2 p.m. call in Bel Air — system condition explanation scored 8. The homeowner asked twice what 'heat exchanger' meant. Today, when you find a heat exchanger issue, try 'the part of the furnace that transfers heat from the burner to your air' and watch what happens." Tech acknowledges; manager moves to the next. No defensiveness, no grades-on-the-fridge dynamic, no public shaming — the framing is "AI helps you see what you couldn't see," not "AI is watching you."

Compliance below 80% on the huddle kills the lift. The mechanism is feedback-loop compression: yesterday's call observed, this morning's specific moment coached, today's reps internalize the fix. Skip the huddle, the loop breaks. One in five days broken means the bad patterns embed during the broken-loop days and leak into closed-loop days. Documented at deploying shops: huddle compliance under 80% maps to a lift of 8-10% instead of 18%. The huddle is not optional. The manager's calendar protects it.

Comp Redesign That Ships at Rollout

The most common Rilla rollout failure is not the tool. It is the comp plan that wasn't redesigned. Pre-Rilla tech comp is typically a flat hourly wage plus a percentage of sold revenue or a SPIF on memberships. The tech accepts being recorded because they have to, but they see more work (recording every call, accepting AI scoring, doing morning huddles) for the same pay. Within 60-90 days, the best techs (most marketable) leave for shops paying volume comp at a less coached environment.

The comp redesign solves this by paying for behavior change before revenue lift shows. The structure: base preserved; variable redesigned to include an RGA bonus tier ($300-$500/month at sustained 75+ RGA, $200/month at sustained 70-75 RGA); a financing-mentioned bonus ($25 per closed ticket where financing was presented and the soft-pull happened); an MPR-attach bonus tier (graduated $50-$200 per closed ticket with membership attached); and a recall reduction bonus (quarterly $500 if the tech's personal recall rate stays under 2%). Total tech pay at full Rilla performance lands +12-18% above pre-Rilla baseline. The economics work because the close-rate lift, MPR lift, and recall reduction produce 3-5x the comp delta in incremental shop margin.

The comp tweak ships at rollout, not after the lift shows. Without it, the tech floor sees more work for same pay and revolts within 60 days. With it, the tech floor sees their personal RGA climbing and their paycheck climbing in lockstep — and the rollout sticks. The conversation that frames the comp redesign is critical: "We are paying you more to do the work the AI helps you do better. The AI is not replacing you; it is making you the highest-paid tech in this market."

Kill Rules and Pilot Discipline at 30, 60, 90 Days

A Rilla pilot in trouble surfaces inside 60 days. The service manager who runs the pilot writes kill rules upfront — the conditions under which the pilot pauses, gets restructured, or rolls back. Without kill rules, a struggling pilot drags for two quarters before anyone has the data to make a clean decision.

Kill rule one: huddle compliance below 80% for two consecutive weeks. Mechanism check — is the manager's calendar blocking the huddle, is the team's stand-up cadence overlapping, is the tech floor pushing back on the time. Fix the mechanism within seven days or pause the pilot for a comp-and-cadence reset.

Kill rule two: RGA team median below 60 at day 45. The tool is producing scores; the techs are not improving. The likely culprits: the coaching framing is feeling punitive (frame check), the comp redesign hasn't shipped (comp check), or the targeted moments are mismatched to the team's actual blind spots (rubric check). Fix one in 14 days or pause and rebuild.

Kill rule three: more than one tech departure during the pilot window attributable to Rilla. Talent loss kills the ROI. Exit interviews surface the cause: comp, framing, recording anxiety, manager dynamic. Address before the next departure.

Kill rule four: close rate decline weeks 4-8. This is the Rilla-induced overcoaching paralysis — techs start performing the coached behaviors mechanically and lose the natural flow that made them good in the first place. The fix is coaching de-intensification (one moment per tech per week, not three), restoring the autonomy that the AI scoring inadvertently undermined.

The day-90 success picture: team RGA median 70+, close rate per tech up 12-18%, MPR up 8-15 points, recall percentage trending toward 2%, no tech departures attributable to the rollout, huddle compliance 90%+. Hit four of five and the pilot graduates to permanent operation. Miss two of five and the pilot restructures.

What the Service Manager Still Owns When AI Scores Every Ride

The service manager's reflex when Rilla lands is the same reflex the CSR and the dispatcher and the advisor have — "the AI is replacing me." The honest answer is the role evolves into the highest-leverage version of itself. The AI does the apprentice work of observing and scoring 30-40 calls per day. The manager does the journeyman work of coaching the human moments AI surfaces.

Five categories of work stay permanently with the service manager. The coaching conversation itself — Rilla flags the moment; only the human delivers the coaching with calibration to the tech's mood, history, and reception bandwidth. The cohort analysis judgment — the data shows coaching-responsive vs. non-responsive; the manager decides which techs get more coaching, which get different coaching, which get a PIP, which get a path off the close-heavy seat. The comp-plan-vs-RGA alignment check — the data surfaces the mismatch; the manager negotiates the comp redesign with the owner. The cross-tech pattern recognition — the AI scores per-tech; the manager sees the team-wide pattern (everyone is weak on financing pivot in August, suggesting a script issue not a tech issue). The protect-the-tech-from-the-AI judgment — when a tech has a personally hard call (sick family member, divorce, the customer was abusive), the manager pulls that call from the scorecard before it depresses the RGA unfairly.

The evolved 2026 service manager runs a 12-tech team with daily ride-along coverage, sustained team RGA 70+, technician close rate at 38-45% on aged-equipment service calls, MPR at 38-45%, recall percentage under 2.5%, and zero tech departures attributable to the AI rollout. The manager's calendar is freed from in-person ride-along days (3 days of windshield time per week pre-Rilla) and reallocated to coaching conversations, cohort analysis, and the owner's Friday recap. The role is more leveraged, not less.

Key Takeaways

• Rilla produces 30-40 virtual ride-alongs per manager per day vs. 2-3 in-person — 10-20x coverage density. $200-$400/seat/month. 18% close-rate lift documented across home-services deployments. Payback in weeks.
• The AI flags five coaching moments per ride: intro (15% weight), system condition explanation (25%), repair-vs-replace pivot (20%), options presentation (15%), financing pivot (25%). Each scored 0-20 or 0-25; aggregate is the RGA.
• RGA is the close-rate leading indicator. Techs at sustained 75+ close at 38-45% on aged equipment. Techs at 55-65 close at 18-25%. 30/60/90 onboarding targets for new hires: 50/65/75 RGA.
• The daily 15-minute morning huddle is the mechanism. Five mornings a week, 2 minutes per tech, frame "AI helps you see what you couldn't see" — not "AI is watching you." Compliance under 80% drops the lift from 18% to 8-10%.
• Comp redesign ships at rollout, not after the lift shows. RGA bonus ($300-$500/mo at 75+), financing-mentioned bonus ($25/closed ticket), MPR-attach bonus ($50-$200), recall-reduction quarterly bonus. Total tech pay +12-18% over pre-Rilla baseline.
• Kill rules at 30/60/90: huddle compliance under 80% for two weeks; team RGA median under 60 at day 45; more than one tech departure attributable to Rilla; close rate decline weeks 4-8 (overcoaching paralysis). Hit four of five success criteria at day 90 or restructure.
• System condition explanation is the highest-impact moment (25% weight). Tech uses homeowner's language, shows AI-tagged photos, ties condition to outcome (safety, comfort, utility bill, recurring repair cost), avoids close-too-early.
• Financing pivot is the most under-coached moment in pre-Rilla shops. Techs default to silence; financing-mentioned calls close at 22-30 points higher. Wisetack for fast service-trade ($1K-$12K), GreenSky for high-ticket ($12K-$50K), Synchrony for branded revolving on multi-system customers.
• Cohort tracking distinguishes coaching-responsive from non-responsive. RGA moves and close rate follows = responsive. RGA moves and close rate flat = mechanical performance without internalization, or wrong moments targeted.
• What the service manager still owns: coaching conversation delivery, cohort analysis judgment, comp-plan-vs-RGA alignment, cross-tech pattern recognition, protect-the-tech-from-AI judgment when calls are personally hard. The role evolves; it does not disappear.